While epidemiological studies have shown the health benefits of physical activity, they also have shown that prevalence of physical activity is well below recommended levels. Exercise- promotion interventions are typically ineffective, and lack of knowledge of exercise determinants inhibits the design of effective interventions. A previous cross-sectional random- sample survey was conducted by this group of investigators to identify correlates of physical activity in adults. The field now needs a multivariate prospective study of the "predictors" of regular physical activity and change in exercise. The proposed behavioral epidemiology study allows for a two-year prospective follow-up of the original survey sample of 1811. High resurvey rates (i.e., greater than 80%) are expected because (a) subject tracking efforts already have been implemented, (b) a significant incentive for responding will be provided, and (c) in a pilot study 95% of respondents agreed to be resurveyed. The Follow-up Survey will elicit a two-year history of "vigorous" exercise and moderate-intensity physical activity (i.e., walking), both of which appear to have substantial health benefits, as well as different determinants. Potential predictors were assessed at the Baseline Survey and will include both demographic variables as well as indices based on Social Learning Theory. Since our cross-sectional analyses indicated that social learning variables accounted for important amounts of variance in exercise, it is likely that they will be confirmed in this prospective analysis. Multiple regression analyses will determine the combined and independent relationships of such variables as self-efficacy, social support, perceived barriers and benefits, and environmental factors on both types of physical activity change. A second analysis strategy will separate subjects into three baseline physical activity groups (i.e., low, intermediate, and high activity) and examine prediction of change versus no-change in each category. There are data to indicate that predictors of adoption of activity by sedentary individuals may be different from relapse from relapse from regular activity by exercisers. Results will aid in understanding change in physical activity, and will be used to generate hypotheses for health promotion interventions. Thus, this study is designed to contribute to meeting the Health Objectives of the Nation.